## Presentation on theme: "Dyadic Analysis: Using HLM"— Presentation transcript:

David A. Kenny University of Connecticut /

Dyads with a categorical within-dyads variables that makes a difference e. g., parent-child Indistinguishable Ordering of the two members is arbitrary e.g. roommates Whether dyads are distinguishable or not is matter of theoretical and statistical considerations.

Nonindependence: Definition
Degree of greater similarity (or dissimilarity) between linked observations versus unlinked observations Nonindependence as the correlation between linked observations.

Negative Nonindependence
Two scores from the same dyad are more dissimilar than two scores from different groups. How?

Compensation: If one person has a large score, the other person lowers his or her score. For example, if one person acts very friendly, the partner may distance him or herself, Social comparison: The members of the dyad use the relative difference on some measure to determine some other variable. For instance, satisfaction after a tennis match is determined by the score of that match. Zero-sum: The sum of two scores is the same for each dyad. For instance, the two members divide a reward that is the same for all dyads. Division of labor: Dyad members assign one member to do one task and the other member to do another. For instance, the amount of housework done in the household may be negatively correlated.

Dyadic Designs Standard Each person has one partner
Social Relations Model (SRM) Designs Each person has many partners, and each partner paired with many persons One-with-Many Each person has many partners, but each partner is paired with only one person

Standard Design

* One-with-Many Design *

* SRM Designs *

Remaining part of the presentation focuses on the standard design which is used in most dyadic studies.

Individual One record for each individual Only that individual’s data on the record Dyad (useful for distinguishable dyads) Each record one dyad Different variables for each person Pairwise (useful for distinguishable dyads) One record for each person The person’s data and partner data included (each data point included twice)

Pairwise Data Organization
Dyad Number Member Number The Person’s Data Partner’s Data Acitelli Example

Both members have the same score X = X′ Within Dyads Sum of two scores a constant X + X′ = c for all dyads

Between

Within

Mixed Variables Variables that vary within and between groups Examples
Most outcomes (and mediators) Individual differences (e.g., personality)

Mixed

Gender and the Three Types
Between Gay and Lesbian Couples Within Heterosexual Married Couples Mixed Friends

Actor-Partner Interdependence Model
X Y partner partner actor X' Y'

Types of APIM Models actor only a > 0; p = 0 partner only
couple model a = p social comparison model a + p = 0

Actor-Partner Interaction
Partner effect depends on the level of the actor effect More than one way to test A Level 2 Variable

Measurement of the Actor-Partner Interaction
Standard Product Approach (XX’) – make sure variables centered Discrepancy Score (│X – X′│) (Absolute Difference) Similarity – Negative Coefficient Dissimilarity – Positive Coefficient Higher Score of the Two [Max(X,X′)] – “It Takes Two” Lower Score [Min(X,X′)] – “One of Us Can Drag Us Both Down”

Estimation Indistinguishable Members Multilevel Modeling
Pairwise Data File Illustrate with HLM Distinguishable Members Structural Equation Modeling Dyad Data File

Approaches Illustrated
Conventional Multilevel Model with Random Intercepts Two-Intercept Model

Briefly Discussed One-with-many design
Standard design studied longitudinally

Similar presentations